Applications such as collaborative mapping motivate a need for network-wide broadcasting in Mobile Ad Hoc Networks (MANET) with low-rate long-range links. Existing MANET protocols are inadequate in such low-capacity regimes due to their high use of control packets. We present a novel protocol called ECHO that constructs and maintains a broadcast backbone without using any control packets. Instead, using a field in the data packet header, a node transmits and listens for an “echo” of the specific packet to determine its membership in the backbone. ECHO is deterministic, source-independent, fully distributed, accommodates mobility, and balances battery consumption across nodes. We prove ECHO’s correctness and show that its communication complexity is lower than that of Multi-Point Relay (MPR) and Flooding, and comparable to that of Opportunistic Announcement (OA). Simulations over random mobile low-capacity networks show that ECHO provides at least a 30 percent better delivery ratio than Flooding, MPR, and OA for 50-node networks and dramatically reduces the communication load. Experiments on a 12-node testbed of goTenna mobile mesh networking devices show that ECHO reduces transmissions by about 3x and increases battery life by more than 50 percent over Flooding. ECHO’s performance advantages are crucial for scalable broadcast in low-power, low-capacity wireless networks.

Most Software-Defined Networking (SDN) based centralized architectures for mobile multi-hop wireless networks propose using one or more of the following: infrastructure for hosting stationary SDNC, out-of-band single-hop (direct) communication links, location-tracking, or existing IP connectivity. Such architectures are inadequate for infrastructure-less networks having low-capacity links and susceptibility to high interference, packet losses, and collisions. We describe a novel centralized opportunistic reactive routing protocol called CORR that allows a mobile SDN Controller (SDNC) to establish connectivity by identifying a subset of nodes in the network and making them the network backbone for forwarding packets. The SDNC learns the network topology via the backbone nodes, reducing the network load and improving the scalability. We evaluate CORR for a wide range of network sizes, densities, and traffic rates. Our ns3 simulation results indicate that CORR provides up to 10% better packet delivery ratio than AODV while incurring at least 1.5x lower network load and 3x lower delay, making it energy efficient and suitable for time-sensitive applications.

Decentralized off-grid short-burst communications for public safety and other applications require mobile wireless networks that can be inexpensively and instantly deployed to cover a large area. Maximizing coverage often necessitates trading bit-rate for range. The resulting capacity is insufficient to support existing mesh networking protocols due to their excessive control overhead, and motivates a fresh approach. We present the networking architecture of goTenna - a long-range lightweight device for mobile mesh networking. Our protocol uses a novel zero-control-packet approach for broadcasting and unicasting that builds state by observing packet header information. We describe its experimental evaluation using ns3, and on a goTenna testbed, concluding with some research challenges.

Traditional network management systems are not usually able to differentiate between mobility and other causes of communication degradation in wireless mobile ad hoc networks. A fault management system needs the ability to not only detect changes in performance but also reason about their possible causes and how to fix the problems. We propose a system called TimeSAFE (Time Series Analyzer Front End) that performs time series analysis as a front end input to a central fault management system. We show how such an analysis can help to distinguish between motion, obstacles, and interference as possible causes of changes in SINR (Signal to Interference and Noise Ratio). We propose three different methods of time series analysis and also methods for dynamic order selection and dynamic window selection. We describe our implementation of TimeSAFE and provide some preliminary results of detecting changes in SINR and identifying their causes.

This dissertation presents an adaptive probing based tool for fault diagnosis in computer networks by addressing the problems of probe station selection and probe selection. We first present algorithms to place probe stations to monitor the network in the presence of various failures in the network. We then present algorithms for probe selection in an adaptive manner to perform fault diagnosis. We present algorithms considering both deterministic as well as non-deterministic environments. We present evaluation of the proposed algorithms through comprehensive simulation studies. The dissertation is available at http://www.cis.udel.edu/~natu/papers/dissertation.pdf.

Probe stations are specially instrumented nodes from where probes can be sent to monitor the network. Probe station locations affect probing efficiency, monitoring capability, and deployment cost. We present probe station selection algorithms and aim to minimize the number of probe stations and make the monitoring robust against failures in a deterministic as well as a non-deterministic environment. We provide experimental evaluation of the proposed algorithms through simulation results.

Probabilistic inference was shown effective in the nondeterministic diagnosis of end-to-end service failures when applied in a centralized management system where the manager possesses a global knowledge of the system structure and state. Since many networks are organized into multiple administrative domains that may be unable to share configuration and state information, these centralized techniques are not applicable to them. This paper proposes a fault localization technique suitable for multidomain networks with hierarchical routing. The proposed technique divides the computational effort and system knowledge among multiple, hierarchically organized managers. Each manager performs fault localization in the domain it manages and requires only the knowledge of its own domain. We show through simulation that the proposed approach not only improves the feasibility of fault localization in multidomain networks, but also increases the effectiveness of probabilistic diagnosis and makes it realizable in networks of considerable size

The bandwidth distribution scheme (BDS) was designed to combine the advantages of the integrated and differentiated services models and to provide support for scalable per-flow quality of service. In recent studies Hnatyshin et al. showed that the variation of the bandwidth distribution scheme called the exact requested bandwidth range BDS (X-BDS), can support per-flow minimum bandwidth guarantees in a scalable manner. The X-BDS achieves per-flow QoS by maintaining aggregate flow requirements in the network core and distributing these requirements as needed. Based on the obtained information the edge nodes determine the fair allocation of available bandwidth among the active flows. This paper introduces an optimization which allows the X-BDS approach to operate seamlessly in the event of network topology changes. The primary challenge of addressing this problem is determining how to correctly update the aggregate flow requirements maintained in the network core. This paper examines two instances of topology changes, link failure and link restoration, and presents an algorithm which enables the routers in the X-BDS network to properly update the aggregate flow requirements in each of these cases. This paper examines the performance of the introduced algorithm through simulation.

This paper investigates an approach to improving the scalability and feasibility of probabilistic fault localization in communication systems by exploiting the domain semantics of computer networks. The proposed technique divides the computational effort and system knowledge among multiple, hierarchically organized managers. Each manager performs fault localization in the domain it manages and requires only the knowledge of its own domain. Since failures propagate among domains, domain managers cooperate with each other to find a consensus explanation of the observed disorder. We show through simulation that the proposed approach increases the effectiveness of probabilistic diagnosis and makes it feasible in networks of considerable size 1.

We apply Bayesian reasoning techniques to perform fault localization in complex communication systems while using dynamic, ambiguous, uncertain, or incorrect information about the system structure and state. We introduce adaptations of two Bayesian reasoning techniques for polytrees, iterative belief updating, and iterative most probable explanation. We show that these approximate schemes can be applied to belief networks of arbitrary shape and overcome the inherent exponential complexity associated with exact Bayesian reasoning. We show through simulation that our approximate schemes are almost optimally accurate, can identify multiple simultaneous faults in an event driven manner, and incorporate both positive and negative information into the reasoning process. We show that fault localization through iterative belief updating is resilient to noise in the observed symptoms and prove that Bayesian reasoning can now be used in practice to provide effective fault localization.

The current best effort approach to quality of service in the Internet can no longer satisfy a diverse variety of customer service requirements, because of which there is a need for alternative strategies. A promising approach for dealing with this problem is a method called the load distribution scheme (LDS) which dynamically adjusts the traffic load at the network boundary based on feedback from the network. In order to fairly share available resources among individual flows, the load distribution scheme relies on a message exchange protocol which in certain cases may cause significant overhead in the system. We examine the issues related to the problem of the message overhead in the LDS, propose solutions to the problem, and evaluate these solutions through simulation in OPNET.

This paper utilizes belief networks to implement fault localization in communication systems taking into account comprehensive information about the system behavior. Most previous work on this subject performs fault localization based solely on the information about malfunctioning system components (i.e., negative symptoms). We show that positive information, i.e., the lack of any disorder in some system components, may be used to improve the accuracy of this process. The technique presented allows lost and spurious symptoms to be incorporated in the analysis. We show through simulation that in a noisy network environment the analysis of lost and spurious symptoms increases the robustness of fault localization with belief networks. We also demonstrate that belief networks yield high accuracy even for approximate probability input data and therefore are a promising model for non-deterministic fault localization.

We present fault localization techniques suitable for diagnosing end-to-end service problems in communication systems with complex topologies. We refine a layered system model that represents relationships between services and functions offered between neighboring protocol layers. In a given layer, an end-to-end service between two hosts may be provided using multiple host-to-host services offered in this layer between two hosts on the end-to-end path. Relationships among end-to-end and host-to-host services form a bipartite probabilistic dependency graph whose structure depends on the network topology in the corresponding protocol layer. When an end-to-end service fails or experiences performance problems it is important to efficiently find the responsible host-to-host services. Finding the most probable explanation (MPE) of the observed symptoms is NP-hard. We propose two fault localization techniques based on Pearl's (1988) iterative algorithms for singly connected belief networks. The probabilistic dependency graph is transformed into a belief network, and then the approximations based on Pearl's algorithms and exact bucket tree elimination algorithm are designed and evaluated through extensive simulation study.

Fault localization is a process of isolating faults responsible for the observable malfunctioning of the managed system. Previously, fault localization efforts concentrated mostly on diagnosing faults related to the availability of network resources in the lowest layers of the protocol stack. This paper focuses on end-to-end service failure diagnosis as a critical step towards multi-layer fault localization in an enterprise environment. By refining a previously proposed modeling technique, we present a universal method of modeling both availability and performance related problems associated with end-to-end services in a non-deterministic fashion. We introduce and evaluate a novel algorithm that allows an event-driven, incremental diagnosis of end-to-end service failures.

Event pattern detection is the one of the major techniques used for event correlation in network and distributed systems management. This paper focuses on the design issues of event pattern specification languages. The discussion is organized around the event operators in existing event languages that we think are problematic and around the temporal specification aspect. Semantic issues are discussed and various languages are investigated. The study has revealed weaknesses of design in semantic appropriateness and completeness of certain event operators, in flexibility of operator usage and timing specification, in the effectiveness and efficiency of expressions, and in readability of the languages. Based on the findings, we propose a new event language called SEL, which attempts to avoid some of these problems. SEL is novel in its negation operator usage, the way followed-by semantics is provided, and how composite event time is determined in the presence of the negation operator. It is comprehensive yet relatively simple and intuitive to use. Expressions written in SEL appear to be very readable and easy to maintain.

SHAMAN is a framework for hierarchical management of networks with SNMP that provides flexible and dynamic network management by permitting distribution of control and management functions over a hierarchical management structure. We have implemented a prototype software package that contains a SHAMAN intermediate manager (IM), a GUI for controlling the IM and for developing applications for it, and an example application of location management in a mobile tactical battlefield network.

In this paper we study an adaptive measurement-based admission control scheme that is suitable for multimedia applications with diversified QoS requirements. The conservativeness of the proposed admission control scheme is self-adjusted based on the measured QoS performance of the existing multimedia flows. Extensive simulations performed on actual multimedia traffic with different combinations of traffic patterns, QoS requirements, and flow lifetimes have shown that the proposed scheme effectively balances the trade-off between high utilization and reliable QoS under different network conditions, hence enables a significant utilization increase compared with existing measurement-based admission control schemes.

A network management scripting application consists of a set of distributed and possibly mobile delegated management scripts, as well as other non-script network management processes, which cooperate with each other to accomplish a given network management task. Such applications require debugging support: firstly, they are distributed applications, and distributed applications are notoriously difficult to debug. More importantly, the network management and the scripting environment exhibit many peculiar characteristics which necessitate special debugging support. The overall issue addressed by this paper is: how can we effectively debug distributed and mobile network management scripting applications? We also present a simple integrated simulation-replay debugger and explain its usage.

In a wide-area network with high-speed links, congestion may lead to poor utilization and the degradation of the service provided to the users. This paper presents a cooperative congestion management scheme in which the intermediate nodes in the vicinity of a congested link share their resources to offload this link before packets are discarded. The scheme uses a progression of proactive control mechanisms on selected traffic bursts: from local rerouting, through upstream rerouting and quenching, to packet discarding. It complements existing open-loop mechanisms by allowing a more liberal packet admission while preserving the underlying switching architecture. The performance evaluation results obtained via simulation show that its implementation reduces the probability of packet discarding by an order of magnitude.

Memnet is a shared-memory local area network under development at the University of Delaware that provides close coupling to the processors of a physically distributed multiprocessor system. The Memnet local network appears as memory in the physical address space of each processor (host) on the network.

Noahnet is an experimental flood local area network with features such as high reliability and high performance. Noahnet uses a randomly connected graph topology with four to five interconnections per node and a flooding protocol to route messages.

Past studies of memory interference in multiprocessor systems have generally assumed that the references of each processor are uniformly distributed among the memory modules. In this paper we develop a model with local referencing, which reflects more closely the behavior of real-life programs. This model is analyzed using Markov chain techniques and expressions are derived for the multiprocessor performance. New expressions are also obtained for the performance in the traditional uniform reference model and are compared with other expressions-available in the literature. Results of a simulation study are given to show the accuracy of the expressions for both models.